The past decade has seen the introduction of several eff ective therapeutic options for patients with primary hepatocellular carcinoma： these include pharmacological therapies （ie， systemic sorafenib and transarterial chemoembolisation with or without beads）， loco-regional therapies （ie， radiofrequency ablation， percutaneous ethanol injection， and cryotherapy）， and surgical therapies（ie， liver resection or replacement）. There have been very few prospective controlled studies comparing diff erent modalities but， on the basis of clinical experience and prospective studies using historical controls， guidelines for the treatment of patients with hepatocellular carcinoma have been developed and largely accepted. Most hepatocellular carcinomas develop on the background of cirrhosis， so both the cause and severity of the cirrhosis must also be considered when planning treatment for the individual. For most patients with advanced cirrhosis and hepatocellular carcinoma， liver transplantation remains the treatment of choice. However， the increasing demand for liver transplants has exceeded the availability of grafts so pretransplant treatment.

options for patients with hepatocellular carcinoma are now considered. The basis for transplant selection can be controversial and diff erent health-care systems have developed solutions for balancing the often competing demands of equity， justice， use， and benefi t in a manner that is open and transparent. Thus， transplantation is available not to all those who might benefi t from the procedure， but for those who would benefi t most. Patients with hepatocellular carcinoma have to compete with patients with decompensated liver disease and others with an unacceptable quality of life because of liver disease （such as chronic encephalopathy）. This situation is exacerbated by the fact that there is， as yet， no eff ective liver support， and thus a paucity of liver donors usually means death for those who are not selected. In the early years of transplantation， survival was greatly decreased by recurrence of cancer， because there were no validated criteria to identify those in whom transplantation was inappropriate. The landmark paper by Mazzaferro and colleagues showed that posttransplantsurvival could be predicted by the number and size of lesions. The so-called Milan criteria were rapidly established as the standard for transplantation and were introduced to clinical practice to provide not only an objective measure to ensure eff ective use of a scarce resource， but also equity of access. Recognition that some patients who marginally exceeded the Milan criteria had a good survival post-transplant led to some centres implementing the criteria more fl exibly and a ebate arose over the number and size of tumours that would preclude transplantation.

The size and number of liver cancers are merely surrogates of the biological behaviour of the cancer： several approaches have been followed to use serological， histological， and molecular characteristics to identify cancers that are likely to recur after transplantation.As yet， the use of molecular signatures to predict cancers that are suitable for transplantation has not been widely adopted， in part because these markers have not been validated and because histological and molecular markers vary both within and between cancers in the same liver. Furthermore， this approach requires assess ment of tissue pretransplant， and， historically， liver biopsy has been contraindicated before curative surgery because of fears of needle-tract spread， although these fears might be overstated. The information in a further study by Mazzaferro and colleagues，reported in this issue of The Lancet Oncology， is undoubtedly of use to the clinician in selecting appropriate candidates， because they give a probability of outcome.

The current UK criteria for selection include the expectation of a more than 50% probability of survival at 5 years； thus these data allow those with hepatocellular carcinoma to compete fairly with those with other indications for liver replacement. However， the conclusions of this study need to be treated with caution in light of the inevitable limitations of the analysis. First， the data for the analysis are taken from selected patients who were transplanted， many of whom exceeded the Milan criteria， and therefore， extrapolation of the fi ndings to those who have not been transplanted might be subject to selection bias. Second， the researchers looked at outcome， but this is aff ected by many factors other than the burden of cancer， such as donor and recipient factors （cardiovascular and renal status）. Third， the researchers based their analysis on data from the explanted liver rather than on pretransplant imaging. This is a potential pitfall because imaging can often understage disease seen in the explant. Finally， the role of pretransplant and post-transplant interventions needs to be considered. Current data suggest that the immunosuppressive agent sirolimus might reduce the rate and consequences of recurrence， but results of large-scale prospective studies are awaited. Pretransplant interventions might change the natural history posttransplant， and indeed several studies have suggested that transarterial chemoembolisation or radiofrequency ablation might reduce post-transplant recurrence. However， although down-staging （using interventions to make tumours fulfi l the Milan criteria） has been advocated， this strategy might be counterintuitive， because the biological behaviour of a hepatocellular carcinoma can remain unchanged. Of note， the length of time before and after any procedure might help identify those tumours that are slowly growing and， therefore， more appropriate for transplant.

Nevertheless， despite these caveats， the new information provided in the study by Mazzaferro and colleagues6 is a major step towards developing a rational and equitable basis for selection of patients for liver transplantation.